CN111083475B - Quantization transformation coefficient management device and encoder suitable for HEVC standard - Google Patents

Quantization transformation coefficient management device and encoder suitable for HEVC standard Download PDF

Info

Publication number
CN111083475B
CN111083475B CN201911417923.6A CN201911417923A CN111083475B CN 111083475 B CN111083475 B CN 111083475B CN 201911417923 A CN201911417923 A CN 201911417923A CN 111083475 B CN111083475 B CN 111083475B
Authority
CN
China
Prior art keywords
quantized transform
transform coefficient
module
transform coefficients
management
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN201911417923.6A
Other languages
Chinese (zh)
Other versions
CN111083475A (en
Inventor
诸悦
陈晓春
刘易华
潘柳含
肖君寿
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shanghai Fullhan Microelectronics Co ltd
Original Assignee
Shanghai Fullhan Microelectronics Co ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Shanghai Fullhan Microelectronics Co ltd filed Critical Shanghai Fullhan Microelectronics Co ltd
Priority to CN201911417923.6A priority Critical patent/CN111083475B/en
Publication of CN111083475A publication Critical patent/CN111083475A/en
Application granted granted Critical
Publication of CN111083475B publication Critical patent/CN111083475B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04NPICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/10Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
    • H04N19/102Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
    • H04N19/124Quantisation
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04NPICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/42Methods or arrangements for coding, decoding, compressing or decompressing digital video signals characterised by implementation details or hardware specially adapted for video compression or decompression, e.g. dedicated software implementation
    • H04N19/423Methods or arrangements for coding, decoding, compressing or decompressing digital video signals characterised by implementation details or hardware specially adapted for video compression or decompression, e.g. dedicated software implementation characterised by memory arrangements
    • H04N19/426Methods or arrangements for coding, decoding, compressing or decompressing digital video signals characterised by implementation details or hardware specially adapted for video compression or decompression, e.g. dedicated software implementation characterised by memory arrangements using memory downsizing methods
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04NPICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/70Methods or arrangements for coding, decoding, compressing or decompressing digital video signals characterised by syntax aspects related to video coding, e.g. related to compression standards
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04NPICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/90Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using coding techniques not provided for in groups H04N19/10-H04N19/85, e.g. fractals
    • H04N19/91Entropy coding, e.g. variable length coding [VLC] or arithmetic coding

Abstract

The present invention provides a quantized transform coefficient management apparatus, including: a first management module, the first management module comprising: the device comprises a first storage unit and a first processing unit. Through the first storage unit and the first processing unit, the first management module can rapidly access and manage part of quantized transform coefficients output by the first management unit and quantized transform coefficients of the first coding unit, only one storage unit is used for realizing data sharing of the quantized transform coefficients of coding units with different sizes, and the situation that coding units with different sizes need independent storage units is avoided, so that the situation that the quantized transform coefficients in a plurality of storage units are selected, moved or discarded ceaselessly is avoided, the compression capacity of the first storage unit is reduced, the hardware overhead is reduced, the operation power consumption of the first storage unit is reduced, and the working efficiency of the quantized transform coefficient management device is improved.

Description

Quantization transformation coefficient management device and encoder suitable for HEVC standard
Technical Field
The invention relates to the technical field of digital video processing, in particular to a quantized transform coefficient management device and an encoder applicable to HEVC standard.
Background
In order to support High definition Video applications, High Efficiency Video Coding (HEVC) is proposed as a general Video Coding standard in conjunction with the Video Coding Team (JVT-VC).
A quadtree decomposition technique is an important coding tool of HEVC, a frame of image is divided into a plurality of tree coding units (CTUs), and a CTU can be divided into a plurality of Coding Units (CUs) according to the quadtree decomposition technique, and the size of the CU generally includes: 4 × 4, 8 × 8, 16 × 16, 32 × 32, and 64 × 64.
Referring to fig. 1, fig. 1 is a schematic diagram of an encoder suitable for HEVC standard in the prior art, in which a quantized transform coefficient management module receives several quantized transform coefficients from one or more quantization modules, the quantized transform coefficients constitute CUs of different sizes, the CUs of different sizes correspond to different storage units inside the quantized transform coefficient management module, so that the quantized transform coefficients are stored in a plurality of independent storage units, and a subsequent quantized transform coefficient management module provides locally or globally optimal quantized transform coefficients to one or more inverse quantization modules and an entropy coding module through mode decision.
Before mode decision, each CU of a size needs to store a quantized transform coefficient corresponding to an inter mode, a fusion mode and an intra mode to ensure that the quantized transform coefficient can be immediately obtained for image reconstruction operation or for generation of an HEVC code stream once any mode is selected. Taking the quantized transform coefficient management module to receive and manage 4 sizes of CUs 8x8/16x16/32x32/64x64 as an example, this means that the quantized transform coefficient management module needs to store 24480 quantized transform coefficients altogether, and considering that the entropy coding module usually lags behind one CTU in the pipeline for hardware implementation, the quantized transform coefficient management module also needs to reserve space for quantized transform coefficients of the global optimal mode selected for the previous CTU, totaling 30624 quantized transform coefficients, calculated with the maximum possible coefficient value of 16 bits, totaling 489984 bits of storage overhead.
When the mode is judged, the quantized transform coefficient management module carries out mode judgment on the quantized transform coefficients according to the RDO mode judgment information of an external mode judgment module so as to provide the stored quantized transform coefficients of the local or global optimal mode to one or more inverse quantization modules for image reconstruction operation, and simultaneously sends the quantized transform coefficients of the global optimal mode to an entropy coding module for generating an HEVC code stream. At present, in order to meet the requirement of parallel search of different depth search paths on corresponding quad-trees, CUs of different sizes are respectively provided with an independent management module and an independent storage unit, so that when the quantized transform coefficient management module receives the RDO mode decision information from the mode decision module to perform mode decision, the quantized transform coefficient is selected, moved or discarded in different storage units in the quantized transform coefficient management module, and repeated data movement wastes not only the bandwidth of the quantized transform coefficient management module, but also the power consumption.
Therefore, before mode decision, the quantized transform coefficient management module needs to store a large number of quantized transform coefficients, so that the storage cost is high; when the mode is judged, the quantized transform coefficient management module needs to select, move or discard the quantized transform coefficient from different internal storage units, and the processing time is long, so that the hardware complexity of the quantized transform coefficient management module is high.
Disclosure of Invention
The invention aims to provide a quantized transform coefficient management device and an encoder applicable to HEVC (high efficiency video coding) standard, so as to solve the problem that the access power consumption of the quantized transform coefficient management device is large when the quantized transform coefficient management device manages the quantized transform coefficients.
To solve the above-described problems, the present invention provides a quantized transform coefficient management apparatus including:
a first management module for managing quantized transform coefficients of a first coding unit output from an external quantization module, the first management module comprising: the first storage unit is used for storing the quantized transform coefficient of the first coding unit; the first processing unit is used for storing management information of quantized transform coefficients of the first encoding unit and controlling the quantized transform coefficients in the first storage unit to be output to the external inverse quantization module and the external entropy encoding module; wherein sizes of the first coding unit include 4 × 4, 8 × 8, 16 × 16, 32 × 32, and 64 × 64.
Optionally, in the quantized transform coefficient management apparatus, the quantized transform coefficient management apparatus further includes: a second management module for managing quantized transform coefficients of a second coding unit output from an external quantization module, the second management module comprising: the second storage unit is used for storing the quantized transform coefficient of the second coding unit; the second processing unit is used for storing management information of the quantized transform coefficient of the second coding unit and controlling the quantized transform coefficient in the second storage unit to be output to an external inverse quantization module and the first management module; wherein the size of the second coding unit includes 4 × 4 and 8 × 8.
Optionally, in the apparatus for managing quantized transform coefficients, the first management module is further configured to manage the quantized transform coefficients output by the second management module; the first storage unit is also used for storing the quantized transform coefficient output by the second management module; the first processing unit is further configured to store management information of the quantized transform coefficients output by the second management module.
Optionally, in the quantized transform coefficient management apparatus, according to the local RDO mode decision information of the external mode decision module, the second processing unit selects a quantized transform coefficient of a local optimal mode from among the quantized transform coefficients of the second encoding unit, and outputs the quantized transform coefficient of the local optimal mode to the external inverse quantization module and the first management module.
Optionally, in the apparatus for managing quantized transform coefficients, according to local RDO mode decision information and global RDO mode decision information of an external mode decision module, the first processing unit selects a quantized transform coefficient of a local optimal mode from among the quantized transform coefficients of the local optimal mode and the quantized transform coefficients of the first encoding unit, and outputs the quantized transform coefficient of the local optimal mode to the external inverse quantization module, and meanwhile, the first processing unit selects a quantized transform coefficient of a global optimal mode from among the quantized transform coefficients of the local optimal mode and the quantized transform coefficients of the first encoding unit, and outputs the quantized transform coefficient of the global optimal mode to an external entropy encoding module.
Optionally, in the apparatus for managing quantized transform coefficients, in the first management module, an embedded compression algorithm is used to compress the quantized transform coefficients of the first encoding unit and the quantized transform coefficients of the local optimal mode, so that the quantized transform coefficients of the first encoding unit and the quantized transform coefficients of the local optimal mode are stored in the first storage unit in a compressed state.
Optionally, in the quantized transform coefficient management apparatus, the first processing unit reads the quantized transform coefficients stored in the first storage unit and the quantized transform coefficients in the local optimal mode, and decompresses the quantized transform coefficients in the local optimal mode so that the external inverse quantization module receives the decompressed quantized transform coefficients in the local optimal mode, and so that the external entropy encoding module receives the decompressed quantized transform coefficients in the global optimal mode.
Optionally, in the apparatus for managing quantized transform coefficients, in the second management module, an embedded compression algorithm is used to compress the quantized transform coefficients of the second encoding unit, so that the quantized transform coefficients of the second encoding unit are stored in the second storage unit in a compressed state.
Optionally, in the quantized transform coefficient management apparatus, the second processing unit reads the quantized transform coefficients stored in the second storage unit, and decompresses the quantized transform coefficients so that the external inverse quantization module and the first management module receive the decompressed quantized transform coefficients in the local optimal mode.
Optionally, in the apparatus for managing quantized transform coefficients, the first storage unit stores the quantized transform coefficients of the first encoding unit and the quantized transform coefficients output by the second management module in a manner of page-by-page allocation, recovery, and single-linked list management.
Optionally, in the apparatus for managing quantized transform coefficients, the second storage unit stores the quantized transform coefficients of the second encoding unit in a manner of page-by-page allocation, recovery, and single-linked list management.
Optionally, in the quantized transform coefficient management apparatus, the quantized transform coefficient management apparatus further includes: an arbiter for coordinating read/write access by the second processing unit to the quantized transform coefficients within the second storage unit and for coordinating read/write access by the first processing unit to the quantized transform coefficients within the first storage unit.
Based on the same inventive concept, the present invention further provides an encoder suitable for the HEVC standard, comprising:
a quantized transform coefficient management apparatus including a first management module for managing quantized transform coefficients of a first coding unit externally input, the first management module including: the first storage unit is used for storing the quantized transform coefficient of the first coding unit; the first processing unit is used for storing management information of quantized transform coefficients of the first encoding unit and controlling the quantized transform coefficients in the first storage unit to be output to the external inverse quantization module and the external entropy encoding module; wherein sizes of the first coding unit include 4 × 4, 8 × 8, 16 × 16, 32 × 32, and 64 × 64.
A quantization module for providing the quantized transform coefficients of the first coding unit to a first management module of the quantized transform coefficient management apparatus;
the inverse quantization module is used for receiving the quantized transform coefficient output by the first management module of the quantized transform coefficient management device; and the number of the first and second groups,
and the entropy coding module is used for receiving the quantized transform coefficient output by the first management module of the quantized transform coefficient management device.
A mode decision module, configured to provide the local RDO mode decision information and the global RDO mode decision information to a first processing unit of the quantized transform coefficient management apparatus.
Optionally, in the encoder adapted to the HEVC standard, the quantized transform coefficient management apparatus further includes: a second management module for managing quantized transform coefficients of a second coding unit output from an external quantization module, the second management module comprising: the second storage unit is used for storing the quantized transform coefficient of the second coding unit; the second processing unit is used for storing management information of the quantized transform coefficient of the second coding unit and controlling the quantized transform coefficient in the second storage unit to be output to an external inverse quantization module and the first management module; wherein the size of the second coding unit includes 4 × 4 and 8 × 8.
Optionally, in the encoder adapted to the HEVC standard, the quantization module is further configured to provide the quantized transform coefficients of the second coding unit to a second management module of the quantized transform coefficient management apparatus.
Optionally, in the encoder adapted to the HEVC standard, the inverse quantization module is further configured to receive a quantized transform coefficient output by a second management module of the quantized transform coefficient management apparatus.
Optionally, in the encoder adapted to the HEVC standard, the mode decision module is further configured to provide local RDO mode decision information to a second processing unit of the quantized transform coefficient management apparatus.
In summary, the present invention provides a quantized transform coefficient management apparatus, comprising: a first management module, the first management module comprising: the device comprises a first storage unit and a first processing unit. Through the first storage unit and the first processing unit, the first management module can rapidly access and manage part of quantized transform coefficients output by the first management unit and quantized transform coefficients of the first coding unit, only one storage unit is used for realizing data sharing of the quantized transform coefficients of coding units with different sizes, and the situation that coding units with different sizes need independent storage units is avoided, so that the situation that the quantized transform coefficients in a plurality of storage units are selected, moved or discarded ceaselessly is avoided, the compression capacity of the first storage unit is reduced, the hardware overhead is reduced, the operation power consumption of the first storage unit is reduced, and the working efficiency of the quantized transform coefficient management device is improved. Further, the present invention also provides an encoder suitable for HEVC standard, including: the device comprises a quantized transform coefficient management device, a quantization module, an inverse quantization module, an entropy coding module and a mode judgment module, wherein the inverse quantization module and the entropy coding module can quickly receive optimized quantized transform coefficients to quickly perform next processing by utilizing the quantized transform coefficient management device, so that the working efficiency of the encoder is improved.
Drawings
Fig. 1 is a schematic diagram of an encoder suitable for the HEVC standard in the prior art;
FIG. 2 is a block diagram of a quantized transform coefficient management apparatus according to an embodiment of the present invention;
fig. 3 is a schematic diagram of an encoder suitable for the HEVC standard according to an embodiment of the present invention;
FIG. 4 is a schematic diagram of a first memory cell according to an embodiment of the present invention;
wherein, 100-quantized transform coefficient management device, 20-first management module, 21-quantized transform coefficient of global optimum mode, 10-second management module, 11-quantized transform coefficient of local optimum mode, 30-arbitrator, 110-second storage unit, 120-second processing unit, 210-first storage unit, 220-first processing unit.
Detailed Description
The following describes a quantized transform coefficient management apparatus and an encoder suitable for the HEVC standard in detail with reference to the accompanying drawings and specific embodiments. The advantages and features of the present invention will become more apparent from the following description. It is to be noted that the drawings are in a very simplified form and are not to precise scale, which is merely for the purpose of facilitating and distinctly claiming the embodiments of the present invention. Further, the structures illustrated in the drawings are often part of actual structures. In particular, the drawings may have different emphasis points and may sometimes be scaled differently.
The present invention provides a quantized transform coefficient management apparatus, referring to fig. 2 and fig. 3, fig. 2 is a block diagram of a quantized transform coefficient management apparatus according to an embodiment of the present invention, fig. 3 is a schematic diagram of an encoder suitable for HEVC standard according to an embodiment of the present invention, where the quantized transform coefficient management apparatus 100 includes: a first management module 20. The first management module 20 is configured to manage the quantized transform coefficients of the first coding unit output by the external quantization module. The first management module 20 includes: a first storage unit 210 and a first processing unit 220, wherein the first storage unit 210 is used for storing the quantized transform coefficients of the first coding unit; the first processing unit 220 is configured to store management information of quantized transform coefficients of the first encoding unit, and to control the quantized transform coefficients in the first storage unit 210 to be output to the external inverse quantization module and the external entropy encoding module; wherein the sizes of the first coding unit include 4 × 4, 8 × 8, 16 × 16, 32 × 32, and 64 × 64, that is, the first management module 20 manages access to quantized transform coefficients of CUs (4 × 4, 8 × 8, 16 × 16, 32 × 32, and 64 × 64) of all sizes. In the invention, the data sharing of the quantized transform coefficients of the coding units with different sizes can be realized by using only one storage unit, and the condition that the coding units with different sizes all need independent storage units is avoided, so that the condition that the quantized transform coefficients in a plurality of storage units are selected, moved or discarded continuously is avoided, the compression capacity of the first storage unit is reduced, the hardware expense is reduced, the operation power consumption of the first storage unit is reduced, and the working efficiency of the quantized transform coefficient management device is improved.
Further, as shown in fig. 2 and 3, the quantized transform coefficient management apparatus may further include: a second management module 10, said second management module 10 being configured to manage quantized transform coefficients of a second coding unit output by an external quantization module, said second management module 10 comprising: a second storage unit 110 and a second processing unit 120, wherein the second storage unit 110 is used for storing the quantized transform coefficients of the second coding unit; the second processing unit 120 is mainly configured to store management information of the quantized transform coefficients of the second encoding unit, and to control the quantized transform coefficients in the second storage unit 110 to be output to an external inverse quantization module and the first management module 20; wherein the sizes of the second coding unit include 4 × 4 and 8 × 8, that is, the second management module 10 manages access to quantized transform coefficients of small-sized CUs (8 × 8 and 4 × 4). In the present invention, the second management module 10 further optimizes the quantized transform coefficient management apparatus, and on the basis of adding the second management module 10, only two storage units (the second storage unit 110 and the first storage unit 210) can be used to implement data sharing of quantized transform coefficients of coding units of different sizes, thereby avoiding a situation that coding units of different sizes all need independent storage units, and further avoiding a situation that quantized transform coefficients in a plurality of storage units are selected, moved or discarded without stopping, thereby reducing operating power consumption of the second storage unit and the first storage unit, and improving working efficiency of the quantized transform coefficient management apparatus.
In this embodiment, the first management module 20 is further configured to manage the quantized transform coefficients output by the second management module; the first storage unit 210 is further configured to store the quantized transform coefficient output by the second management module 10; the first processing unit 220 is further configured to store management information of the quantized transform coefficients output by the second management module 10.
In this embodiment, according to the local RDO mode decision information of the external mode decision module, the second processing unit 120 selects the quantized transform coefficient 11 of the local optimal mode from the quantized transform coefficients of the second encoding unit, and outputs the quantized transform coefficient 11 of the local optimal mode to the external inverse quantization module and the first management module 20. When the quantized transform coefficient 11 of the local optimum mode is output, the quantized transform coefficient in the second storage unit 110, which is the same as the quantized transform coefficient 11 of the local optimum mode, is released. For example, if the inter mode of an 8 × 8 CU is selected as the locally optimal quantized transform coefficient 11, the locally optimal quantized transform coefficient 11 (including luminance and chrominance coefficients) is sent to the external inverse quantization module and the first management module 20, and the same CU as the locally optimal CU in the second storage unit 110, the quantized transform coefficients included in the CU, and the quantized transform coefficients included in all lower layer CUs are released.
Further, according to the local RDO mode decision information and the global RDO mode decision information of the external mode decision module, the first processing unit 220 selects the quantized transform coefficient 11 of the local optimal mode from the quantized transform coefficient 11 of the local optimal mode and the quantized transform coefficient of the first encoding unit, and outputs the quantized transform coefficient 11 of the local optimal mode to the external inverse quantization module, and meanwhile, the first processing unit 220 selects the quantized transform coefficient 21 of the global optimal mode from the quantized transform coefficient of the local optimal mode and the quantized transform coefficient of the first encoding unit, and outputs the quantized transform coefficient 21 of the global optimal mode to the external entropy encoding module.
Preferably, in the first management module, the quantized transform coefficient of the first coding unit and the quantized transform coefficient 11 of the local optimal mode are compressed by using an embedded compression algorithm, so that the quantized transform coefficient of the first coding unit and the quantized transform coefficient 11 of the local optimal mode are stored in the first storage unit 210 in a compressed state. Similarly, in the second management module 10, the quantized transform coefficient of the second coding unit is compressed by using an embedded compression algorithm, so that the quantized transform coefficient of the second coding unit is stored in the second storage unit 110 in a compressed state. In addition, the second storage unit 110 may be managed in units of 256 bits, which is a 16-coefficient unit, or the embedded compression algorithm may not be used, and the second management module 10 may manage the space of the second storage unit 110 using a direct mapping table, that is, one 16-coefficient space is allocated from the second storage unit 110 to store data when quantized transform coefficients are input. The invention does not limit whether the second management module adopts the embedded compression algorithm. Specifically, the present embodiment takes CU with a basic unit of 4x4 as an example, and specifically introduces the embedded compression algorithm, and the present invention uses a dual-mode embedded compression algorithm, where 16 quantized transform coefficients, denoted as c, are shared by CU with 4x4i. Each quantized transform coefficient is 16 bits, where c0For quantizing the transform system for DCAnd (4) counting. 16 c before compressioniThe total length is 256 bits. First of all, the quantized transform coefficient ciNeeds to be transformed to obtain ci' the specific transformation formula is as follows:
Figure BDA0002351661900000081
then, c is calculatediEffective bit width of `, calculate ciThe formula for the effective bit width of' is as follows:
Figure BDA0002351661900000091
wherein the content of the first and second substances,
Figure BDA0002351661900000092
then, c is added0Bit truncation to bit width bw + 1; c is toi′,i∈[0,15]Truncate to bit width bw. Referring to table 1, in Mode ═ 0, c after truncationi' stored in turn in a coeff field, ciThe symbols are sequentially stored in sign fields, and all the fields are continuously stored. Wherein, the flag CBF indicates whether the current block is all 0, and if CBF ═ 1, all the remaining fields are skipped. Further, referring to table 2, in Mode ═ 1, SCF (16bit) denotes ciIs not equal to 0, and no more c equal to 0 is stored in the coeff and sign fieldsiCorresponding symbol and ci', the remaining sign and coeff information are stored continuously.
Mode-0 quantized transform coefficient compressed storage data structure
TABLE 1
CBF Mode bw sign coeff
Mode-1 quantized transform coefficient compressed storage data structure
TABLE 2
CBF Mode bw SCF sign coeff
The inventors have tested that mode 1 has a significant advantage in block efficiency for near all-zeros, but the advantage is not significant when there are more non-zero quantized transform coefficients because the SCF takes up additional space. Therefore, the two modes are tried during compression in the invention, and the mode with smaller capacity is selected as the compression mode which is finally selected. The finally selected compressed Mode is saved as the Mode field immediately after the CBF.
In this embodiment, the second processing unit 120 reads the quantized transform coefficients stored in the second storage unit 110, and decompresses the quantized transform coefficients so that the external inverse quantization module and the first management module receive the decompressed quantized transform coefficients 11 in the local optimal mode; similarly, the first processing unit 220 reads the quantized transform coefficients stored in the first storage unit and the quantized transform coefficients of the local optimal mode, and decompresses the quantized transform coefficients so that the external inverse quantization module receives the decompressed quantized transform coefficients 11 of the local optimal mode, and decompresses the quantized transform coefficients so that the external entropy encoding module receives the decompressed quantized transform coefficients 21 of the global optimal mode. Specifically, the decompression process is the inverse of compression, and since the compressed length can be estimated by bw, it is not necessary to additionally store length information or use separators to divide multiple coefficient blocks. Tests show that the embedded compression algorithm can obtain good effect in a typical application scene, and can complete the coding and decoding work of two 16-coefficient blocks in a single period, so that the application with high throughput rate can be met. Compressing the quantized transform coefficient of the second encoding unit in the second storage unit 110 using the embedded compression algorithm, and compressing the quantized transform coefficient of the first encoding unit and the quantized transform coefficient 11 of the locally optimal mode in the first storage unit 210, greatly reduces the compressed storage capacity of the first management module 20, thereby reducing the storage overhead of the quantized transform coefficient management module 100, and thus reducing the hardware overhead of the system.
Further, the first storage unit 210 includes a plurality of single-port RAMs, for example, 4 single-port RAMs of 384 × 128 bits are used to store quantized transform coefficients amounting to no more than 192 kbit. The RAM space is divided into 768 nodes, each node has 256 bytes, and the nodes are interleaved and stored in 4 RAMs to provide 512 bits of peak access bandwidth per period. Due to the application of the embedded compression algorithm, it is sufficient to meet the bandwidth requirements of multiple simultaneous accesses. The access management uses a linked list to map the logical address of each mode (fusion mode, inter-frame mode, and intra-frame mode) to a physical address, the first processing unit 220 includes at least one dual-port RAM, and the dual-port RAM is used to temporarily store management information for managing the quantized transform coefficients of the first encoding unit, where 768 × 10 bits of linked list pointers are stored in the dual-port RAM, and each mode of each CU has a separate linked list head-tail pointer management linked list.
Preferably, the second storage unit 110 stores the quantized transform coefficients of the second coding unit in a manner of page-wise allocation, recovery and single-linked list management; the first storage unit 210 stores the quantized transform coefficients of the first encoding unit and the quantized transform coefficients output by the second management module 10 in a page-by-page allocation, recovery, and singly linked list management manner. In this embodiment, the second storage unit 110 includes a plurality of single-port RAMs, and the second storage unit 110 can store quantized transform coefficients of 8 × 8 CUs and 4 × 4 CUs, for which mode decision has not been completed, and it is seen that the capacity of the second storage unit 110 is very small. The second storage unit 110 uses a single-port RAM with a 256-bit width, and because reading is very little, no obvious read-write conflict problem is caused.
Wherein, in addition to receiving the quantized transform coefficients 11 of the locally optimal mode of the second management module 10, the first management module 20 compresses all the input quantized transform coefficients in units of 16 coefficients in the order in which the quantization module outside outputs the quantized transform coefficients and stores them in the first storage unit 220 inside, and in the present invention, the order in which the quantization module outside outputs the quantized transform coefficients is a transposed raster scan order, and the use of a transposed raster order instead of 4 × 4 block compression slightly reduces compression efficiency but saves the conversion operation of the quantized transform coefficients into and out of the integer transform module outside and the integer inverse transform outside. In this embodiment, the first storage unit 210 is used to store the logically continuous code stream continuously, the physical data is managed by using the single chain table, and when a single node of the single chain table is not stored, the subsequent quantized transform coefficient is stored in the subsequent node of the link continuously, so that the physical addresses in the single-port RAM are not necessarily continuous. The non-partition modes of each CU are independently allocated with addresses and stored in the RAM, and the CUs, namely subtrees of the quad-tree, forming the partition modes are also independently allocated with addresses and stored. In a specific example of the present invention, quantized transform coefficients of each of 4x4 CUs and 8x8 CUs included in a 16x16 CU partition are stored continuously. Since the storage capacity occupied by each 16 coefficients can be calculated during decompression, no segmenter needs to be added. The embedded compression algorithm of the present invention uses variable length coding, so that it is necessary to align the first storage unit 210 to a 128-bit word length before writing into it, and therefore, CUs of different sizes each have a 128-bit splicing register for temporarily storing the quantization transform coefficients that have not been written into.
Referring to fig. 4, fig. 4 is a schematic diagram of the operation of the first storage unit according to the embodiment of the present invention, when the quantized transform coefficients of the 16 × 16 local optimum mode from the second management module 10 and the quantized transform coefficients from the external quantization module are input into the first storage unit 210, the input quantized transform coefficients a are compressed into a in units of 16 coefficients, and are spliced with the quantized transform coefficients D stored in the 128-bit splicing register corresponding to the mode to form new data D'. At this time, the first storage unit 210 performs the following steps: the first step is as follows: if len (D') > 128 or a is the last 16 coefficients in the CU, then the second step is performed, otherwise the next 16 coefficient inputs are waited for at the current position; wherein len () is a length-taking operation; the second step is as follows: (1) checking the linked list corresponding to the mode corresponding to the a, if the linked list is empty, applying for an idle node and adding the idle node into the linked list; (2) when c is the last 16 coefficients in the CU, then all D ' are written into the first storage unit 210 and the splicing register where D ' is located is emptied, otherwise, the front part of D ' is used
Figure BDA0002351661900000111
Write to the tail node of the linked list and leave the remaining bits in the splice register. Wherein the content of the first and second substances,
Figure BDA0002351661900000112
to round-down operations. And if the capacity of the chain table tail node is full in the writing process, applying for an idle node and adding the idle node into the chain table tail, and writing the residual data into the newly added idle node.
Further, when the mode decision module outside is used to perform RDO mode decision on a CU of a certain size in the first management module 20, and select a quantized transform coefficient 21 in a global optimal mode (e.g. inter mode) from a partition mode and 3 non-partition modes, the first management module performs the following steps: the storage space occupied by the remaining other modes (e.g., the merge mode and the intra mode) of the same CU is released, i.e., it is sequentially connected to the tail of the free chain, and its linked list is set to be the empty chain. If the global optimal mode is a non-partition mode, reading the content of each node item by item from the head of a chain table of a CU in the mode, sequentially compressing and sending the quantized transform coefficients obtained by decompression to the inverse quantization module for reconstruction; and if the current mode is not the root node of the CTU, adding the chain table of the global optimal mode CU to the chain tail of the partition mode of the upper layer CU, and otherwise, waiting for the chain table of the previous CTU global optimal mode to be empty and pointing the chain table to the current CU mode chain table. When the linked list of the global optimal mode of the previous CTU is not empty, executing the step (1): and reading the data of the first node, decompressing and sending the data to the entropy coding module for entropy coding. Deleting the first node from the chain of the global optimal mode of the previous CTU, and putting the first node back to the idle chain; step (2): and (3) repeating the step (1) until the global optimal mode linked list of the previous CTU is empty.
Further, when applying for an idle node, executing the step (1): if the idle linked list is empty, deleting a chain corresponding to the non-split mode to be selected, and adding the chain to the idle linked list, wherein the selection of the chain corresponding to the non-split mode to be selected can select a mode which is most probably selected by the RDO mode decision according to a rapid mode decision scheme such as SATD (transformation difference absolute sum) and the like, and can also select one mode optionally, and tests show that no detectable performance difference exists under reasonable parameters. The subsequent operation can have two schemes: the RDO algorithm may be set to disable the selection of the mode corresponding to the deleted chain or to regenerate the quantized transform coefficients if this mode is selected by the RDO. The former will suffer a slight loss in coding performance and the latter will suffer a slight loss in throughput. However, the actual test result shows that the idle linked list is almost impossible to be empty in the actual scene under reasonable parameters, so that the loss of the two schemes can be ignored in the actual scene, and the two schemes are not essentially different; and (2) deleting the first node from the idle linked list, and returning the first node as a result of applying for returning.
In this embodiment, in the first management module 20, the three non-partition modes of each CU are sorted from large to small according to SATD (absolute sum of transformed differences) of each mode before RDO mode decision, when the RAM applies for an idle node and the idle node is full, the modes are sequentially discarded in the order of large to small CU size and large to small SATD until the RAM is idle, and the discarded mode is also discarded in the RDO mode decision to prevent errors, but the partition mode is not discarded due to insufficient capacity. The test result shows that the storage overhead of the first storage unit 210 is reduced to 40% of the standard design, in the invention, the first storage unit 210 is used for accessing the quantized transform coefficient, the first storage unit 210 avoids the situation of selecting, moving or discarding the quantized transform coefficient for multiple times, the reading time is shortened, and the power consumption of the quantized transform coefficient management device is reduced.
Further, the quantized transform coefficient management apparatus 100 further includes: an arbiter 30, said arbiter 30 being adapted to coordinate read/write access of said quantized transform coefficients in said first storage unit 210 by said first processing unit 220 and to coordinate read/write access of said quantized transform coefficients in said second storage unit 110 by said second processing unit 120.
Based on the same inventive concept, the present invention further provides an encoder suitable for the HEVC standard, and referring to fig. 3, fig. 3 is a schematic diagram of an encoder suitable for the HEVC standard according to an embodiment of the present invention, where the encoder suitable for the HEVC standard includes: a quantized transform coefficient management apparatus 100, a quantization module, an inverse quantization module, an entropy coding module, and a mode decision module, the quantized transform coefficient management apparatus 100 including a first management module 20, the first management module 20 for managing quantized transform coefficients of a first coding unit externally input, the first management module 20 including: a first storage unit 210 and a first processing unit 220, wherein the first storage unit 210 is used for storing the quantized transform coefficients of the first coding unit; the first processing unit 220 is configured to store management information of quantized transform coefficients of the first encoding unit, and to control the quantized transform coefficients in the first storage unit 210 to be output to the external inverse quantization module and the external entropy encoding module; the mode decision module is configured to provide the local RDO mode decision information and the global RDO mode decision information to the first processing unit 220 of the quantized transform coefficient management apparatus; wherein sizes of the first coding unit include 4 × 4, 8 × 8, 16 × 16, 32 × 32, and 64 × 64.
Further, the quantization module is configured to provide the quantized transform coefficients of the first coding unit to a first management module of the quantized transform coefficient management apparatus; the inverse quantization module is used for receiving the quantized transform coefficient output by the first management module of the quantized transform coefficient management device; the entropy coding module is used for receiving the quantized transform coefficient output by the first management module of the quantized transform coefficient management device.
In this embodiment, the quantized transform coefficient management apparatus 100 further includes: a second management module 10, configured to manage quantized transform coefficients of a second coding unit output by an external quantization module, where the second management module 10 includes: a second storage unit 110 and a second processing unit 120, wherein the second storage unit 110 is used for storing the quantized transform coefficients of the second coding unit; the second processing unit 120 is configured to store management information of the quantized transform coefficients of the second encoding unit, and to control the quantized transform coefficients in the second storage unit 110 to be output to an external inverse quantization module and the first management module 20; wherein the size of the second coding unit includes 4 × 4 and 8 × 8. Furthermore, the quantization module is further configured to provide the quantized transform coefficients of the second coding unit to a second management module 10 of the quantized transform coefficient management apparatus; the inverse quantization module is further configured to receive the quantized transform coefficient output by the second management module of the quantized transform coefficient management apparatus.
Further, the mode decision module is further configured to provide the second processing unit 120 of the quantized transform coefficient management apparatus 100 with local RDO mode decision information, respectively. On the one hand, according to the local RDO mode decision information of the external mode decision module, the second processing unit 120 selects the quantized transform coefficient 11 of the local optimal mode from the quantized transform coefficients of the second encoding unit, and outputs the quantized transform coefficient 11 of the local optimal mode to the external inverse quantization module and the first management module 20. On the other hand, according to the local RDO mode decision information and the global RDO mode decision information of the external mode decision module, the first processing unit 220 selects the quantized transform coefficient 11 of the local optimal mode from the quantized transform coefficient 11 of the local optimal mode and the quantized transform coefficient of the first encoding unit, and outputs the quantized transform coefficient 21 of the global optimal mode to the external inverse quantization module; meanwhile, the first processing unit 220 selects a quantized transform coefficient 21 of a global optimum mode from the quantized transform coefficient 11 of the local optimum mode and the quantized transform coefficient of the first encoding unit, and outputs the quantized transform coefficient 21 of the global optimum mode to an external entropy encoding module. In the present invention, with the quantized transform coefficient management apparatus 100, the dequantization module can quickly receive the quantized transform coefficient 11 in the local optimal mode to perform the image reconstruction operation, thereby improving the work efficiency of the dequantization module, and the entropy coding module can quickly receive the quantized transform coefficient 21 in the global optimal mode to quickly generate the HEVC code stream, thereby improving the work efficiency of the entropy coding module, thereby improving the work efficiency of the encoder.
In summary, the present invention provides a quantized transform coefficient management apparatus, comprising: a first management module, the first management module comprising: the device comprises a first storage unit and a first processing unit. Through the first storage unit and the first processing unit, the first management module can rapidly access and manage part of quantized transform coefficients output by the first management unit and quantized transform coefficients of the first coding unit, only one storage unit is used for realizing data sharing of the quantized transform coefficients of coding units with different sizes, and the situation that coding units with different sizes need independent storage units is avoided, so that the situation that the quantized transform coefficients in a plurality of storage units are selected, moved or discarded ceaselessly is avoided, the compression capacity of the first storage unit is reduced, the hardware overhead is reduced, the operation power consumption of the first storage unit is reduced, and the working efficiency of the quantized transform coefficient management device is improved. Further, the present invention also provides an encoder suitable for HEVC standard, including: the device comprises a quantized transform coefficient management device, a quantization module, an inverse quantization module, an entropy coding module and a mode judgment module, wherein the inverse quantization module and the entropy coding module can quickly receive optimized quantized transform coefficients to quickly perform next processing by utilizing the quantized transform coefficient management device, so that the working efficiency of the encoder is improved.
The above description is only for the purpose of describing the preferred embodiments of the present invention, and is not intended to limit the scope of the present invention, and any variations and modifications made by those skilled in the art based on the above disclosure are within the scope of the appended claims.

Claims (15)

1. A quantized transform coefficient management apparatus, comprising:
a first management module for managing quantized transform coefficients of a first coding unit output from an external quantization module, the first management module comprising: the first storage unit is used for storing the quantized transform coefficient of the first coding unit; the first processing unit is used for storing management information of the quantized transform coefficient of the first coding unit and controlling the quantized transform coefficient in the first storage unit to be output to an external inverse quantization module and an external entropy coding module; wherein sizes of the first coding unit include 4 × 4, 8 × 8, 16 × 16, 32 × 32, and 64 × 64;
the quantized transform coefficient management apparatus further includes: a second management module for managing quantized transform coefficients of a second coding unit output from an external quantization module, the second management module comprising: the second storage unit is used for storing the quantized transform coefficient of the second coding unit; the second processing unit is used for storing management information of the quantized transform coefficient of the second coding unit and controlling the quantized transform coefficient in the second storage unit to be output to an external inverse quantization module and the first management module; wherein the size of the second coding unit includes 4 × 4 and 8 × 8.
2. The quantized transform coefficient management apparatus of claim 1, wherein said first management module is further configured to manage the quantized transform coefficient output by said second management module; the first storage unit is also used for storing the quantized transform coefficient output by the second management module; the first processing unit is further configured to store management information of the quantized transform coefficients output by the second management module.
3. The quantized transform coefficient management device according to claim 2, wherein said second processing unit selects quantized transform coefficients of a local optimum mode from among the quantized transform coefficients of said second encoding unit, based on local RDO mode decision information of an external mode decision module, and outputs the quantized transform coefficients of the local optimum mode to an external inverse quantization module and said first management module.
4. The quantized transform coefficient management device according to claim 3, wherein said first processing unit selects a quantized transform coefficient of a local optimum mode from among the quantized transform coefficients of the local optimum mode and the quantized transform coefficients of said first encoding unit, and outputs the quantized transform coefficient of the local optimum mode to said inverse quantization module outside, according to local RDO mode decision information and global RDO mode decision information of an external mode decision module, and said first processing unit selects a quantized transform coefficient of a global optimum mode from among the quantized transform coefficients of the local optimum mode and the quantized transform coefficients of said first encoding unit, and outputs the quantized transform coefficient of the global optimum mode to an external entropy encoding module.
5. The apparatus according to claim 4, wherein in the first management module, the quantized transform coefficients of the first coding unit and the quantized transform coefficients of the local optimal mode are compressed by an embedded compression algorithm, so that the quantized transform coefficients of the first coding unit and the quantized transform coefficients of the local optimal mode are stored in the first storage unit in a compressed state.
6. The apparatus according to claim 5, wherein said first processing means reads the quantized transform coefficients stored in said first storage means and the quantized transform coefficients of said locally optimal mode, and decompresses the quantized transform coefficients so that said external dequantization module receives the decompressed quantized transform coefficients of said locally optimal mode, and so that said external entropy encoding module receives the decompressed quantized transform coefficients of said globally optimal mode.
7. The apparatus according to claim 4, wherein in the second management module, the quantized transform coefficients of the second coding unit are compressed by an embedded compression algorithm, so that the quantized transform coefficients of the second coding unit are stored in the second storage unit in a compressed state.
8. The quantized transform coefficient management apparatus according to claim 7, wherein said second processing unit reads the quantized transform coefficients stored in said second storage unit, and decompresses them so that said inverse quantization module and said first management module outside receive the decompressed quantized transform coefficients of said locally optimal mode.
9. The apparatus according to claim 2, wherein said first storage means stores the quantized transform coefficients of said first encoding means and the quantized transform coefficients outputted from said second management means in a manner of page-by-page allocation, reclamation, and single-linked list management.
10. The apparatus according to claim 2, wherein said second storage means stores the quantized transform coefficients of said second encoding means in a manner of page-wise allocation, reclamation, and singly-linked list management.
11. The quantized transform coefficient management device according to claim 2, further comprising: an arbiter for coordinating read/write access by the second processing unit to the quantized transform coefficients within the second storage unit and for coordinating read/write access by the first processing unit to the quantized transform coefficients within the first storage unit.
12. An encoder adapted for use in the HEVC standard, comprising:
a quantized transform coefficient management apparatus including a first management module for managing quantized transform coefficients of a first coding unit externally input, the first management module including: the first storage unit is used for storing the quantized transform coefficient of the first coding unit; the first processing unit is used for storing management information of the quantized transform coefficient of the first coding unit and controlling the quantized transform coefficient in the first storage unit to be output to an external inverse quantization module and an external entropy coding module; wherein sizes of the first coding unit include 4 × 4, 8 × 8, 16 × 16, 32 × 32, and 64 × 64;
a quantization module for providing the quantized transform coefficients of the first coding unit to a first management module of the quantized transform coefficient management apparatus;
the inverse quantization module is used for receiving the quantized transform coefficient output by the first management module of the quantized transform coefficient management device; and the number of the first and second groups,
an entropy coding module for receiving the quantized transform coefficients output by the first management module of the quantized transform coefficient management apparatus;
a mode decision module for providing local RDO mode decision information and global RDO mode decision information to a first processing unit of the quantized transform coefficient management apparatus;
the quantized transform coefficient management apparatus further includes: a second management module for managing quantized transform coefficients of a second coding unit output from an external quantization module, the second management module comprising: the second storage unit is used for storing the quantized transform coefficient of the second coding unit; the second processing unit is used for storing management information of the quantized transform coefficient of the second coding unit and controlling the quantized transform coefficient in the second storage unit to be output to an external inverse quantization module and the first management module; wherein the size of the second coding unit includes 4 × 4 and 8 × 8.
13. Encoder suitable for HEVC standard according to claim 12 wherein the quantization module is further adapted to provide the quantized transform coefficients of the second coding unit to a second management module of the quantized transform coefficient management apparatus.
14. An encoder as claimed in claim 12, adapted for use in the HEVC standard, wherein the inverse quantization module is further adapted to receive the quantized transform coefficients output by the second management module of the quantized transform coefficient management apparatus.
15. Encoder suitable for the HEVC standard as claimed in claim 12, characterized in that said mode decision module is further adapted to provide local RDO mode decision information to a second processing unit of said quantized transform coefficient management means.
CN201911417923.6A 2019-12-31 2019-12-31 Quantization transformation coefficient management device and encoder suitable for HEVC standard Active CN111083475B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201911417923.6A CN111083475B (en) 2019-12-31 2019-12-31 Quantization transformation coefficient management device and encoder suitable for HEVC standard

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201911417923.6A CN111083475B (en) 2019-12-31 2019-12-31 Quantization transformation coefficient management device and encoder suitable for HEVC standard

Publications (2)

Publication Number Publication Date
CN111083475A CN111083475A (en) 2020-04-28
CN111083475B true CN111083475B (en) 2022-04-01

Family

ID=70321023

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201911417923.6A Active CN111083475B (en) 2019-12-31 2019-12-31 Quantization transformation coefficient management device and encoder suitable for HEVC standard

Country Status (1)

Country Link
CN (1) CN111083475B (en)

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1620816A (en) * 2002-01-22 2005-05-25 诺基亚有限公司 Coding transform system in image/video encoders and/or decoders
CN103975592A (en) * 2011-12-15 2014-08-06 联发科技(新加坡)私人有限公司 Method and apparatus for dequantization of transformed coefficients
WO2019126307A1 (en) * 2017-12-19 2019-06-27 Qualcomm Incorporated Quantization parameter control for video coding with joined pixel/transform based quantization
WO2019188097A1 (en) * 2018-03-28 2019-10-03 ソニー株式会社 Image processing device and image processing method

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1620816A (en) * 2002-01-22 2005-05-25 诺基亚有限公司 Coding transform system in image/video encoders and/or decoders
CN103975592A (en) * 2011-12-15 2014-08-06 联发科技(新加坡)私人有限公司 Method and apparatus for dequantization of transformed coefficients
WO2019126307A1 (en) * 2017-12-19 2019-06-27 Qualcomm Incorporated Quantization parameter control for video coding with joined pixel/transform based quantization
WO2019188097A1 (en) * 2018-03-28 2019-10-03 ソニー株式会社 Image processing device and image processing method

Also Published As

Publication number Publication date
CN111083475A (en) 2020-04-28

Similar Documents

Publication Publication Date Title
US7106911B2 (en) Image processing apparatus and control method for inputting image data and encoding the data
TWI514779B (en) Entropy coding
US7129860B2 (en) System and method for performing scalable embedded parallel data decompression
CN101527849B (en) Storing system of integrated video decoder
CN101248430B (en) Transpose buffering for video processing
US6292114B1 (en) Efficient memory mapping of a huffman coded list suitable for bit-serial decoding
JP3220598B2 (en) Variable length code table and variable length coding device
WO2015078422A1 (en) Image encoding and decoding method and device
US20030043905A1 (en) Image processing method and apparatus, computer program, and storage medium
US11139827B2 (en) Conditional transcoding for encoded data
CN109600618A (en) Video-frequency compression method, decompression method, device, terminal and medium
US7800519B2 (en) Method and apparatus for compressing and decompressing data
US20140086309A1 (en) Method and device for encoding and decoding an image
US20120033738A1 (en) Virtual frame buffer system and method
US20150277776A1 (en) Data storage control apparatus and data storage control method
CN105578190A (en) Lossless compression method and system for video hard decoding
CN103888777A (en) Video image compression/decompression device
CN113170140A (en) Bit plane encoding of data arrays
CN101511025B (en) Method and apparatus for compressing and decompressing image
JP4209631B2 (en) Encoding device, decoding device, and compression / decompression system
CN103974090B (en) Image encoding apparatus
US20090037799A1 (en) Operating method applied to low density parity check (ldpc) decoder and circuit thereof
CN1964495A (en) A method and device for carrying out buffer control to real time digital video stream
DE102011100936A1 (en) Techniques for storing and retrieving pixel data
CN111083475B (en) Quantization transformation coefficient management device and encoder suitable for HEVC standard

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant